(Bloomberg) – Nvidia Corp., whose products have driven a flood of artificial intelligence spending, said a new type of AI model that produces more complex answers will only increase the need for infrastructure computing.
Most of them read from Bloomberg
CEO Jensen Huang said the concerns raised by Chinese startup Deekseek’s R1 AI model (there are fewer chips and powerful servers needed for such software in the future) are misguided.
“Understanding the R1 was completely wrong,” fans said at a meeting with analysts and investors at his company’s GTC conference in San Jose, California. “The calculation demand is much higher.”
Nvidia’s shares rose 2.4% in New York trading. They had fallen 14% towards the session this year.
The company is trying to convince the wider industrial group to invest in AI equipment, and promises that the economic benefits of the technology are at hand. Nvidia has become the world’s most valuable chipmaker after eye-opening growth in the last two years, facing investor scrutiny in 2025. The main concern is whether customers will maintain their spending on AI infrastructure.
Deepseek surprised these fears when it released the AI ​​model earlier this year, saying it produced powerful technology for less money. However, Nvidia’s biggest clients have since reaffirmed their spending plans. This week’s analysis by Bloomberg Intelligence shows that spending by the largest data center operators is actually climbing faster than expected.
At the analyst meeting, Nvidia’s Huang was asked about the customer’s efforts to develop their own components. This is a chip that could replace the AI ​​accelerator in your data center. Companies such as Google at Alphabet Inc. have collaborated with Broadcom Inc. to develop unique application-specific integrated circuits or ASICs in this area. However, Huang retorted that while many ASICs are designed, they are not necessarily actually available in data centers.
These large customers need good tips to generate more revenue from their infrastructure, not cheap to save money.
“All of these companies are run by great CEOs who are really good at math,” he said. “The effect is not just a cost. It’s a different calculation.”
Competitor chips cannot match Nvidia’s hopper designs. And the current Blackwell platform is 40 times more powerful.
